rag_with_judge

RAG API web server with confidence judging

Similar to rag_api except that it first lets the LLM attempt to answer the query. If not confident enough to answer, re-runs the query using data retrieved from nyx subscriptions.

Overview

• See rag_api
• Uses the Utils.with_confidence prompt modification to demand the LLM judge its own answer.

Setup

See parent README

ℹ️ Make sure you have set the OPENAI_API_KEY environment variable.

Usage

• See rag_api